Use > Service Automation Visualizer > Significant scan result difference heuristics

Significant scan result difference heuristics

When you compare scan results in SAV, specific objects and their attributes are evaluated between each scan result and any differences (and non-existence of objects) is displayed in the comparison results. (For information on the standard set of objects and attributes compared in a scan results comparison, see Compare snapshots.)

In addition to the basic set of attributes evaluated in a comparison, SAV also applies a certain set of heuristics to some attributes in order to discover unique differences that SAV has determined to be interesting or useful.

Specifically, if an attribute value in one of the scan results exceeds a minimum (or maximum) threshold and its value changes by at least a certain percentage between the scan results, then SAV presents this in the comparison results.

The following special object attribute differences are compared:

  • Server: Load average, percentage of free memory on a server, percentage of free swap memory.
  • Server’s File System: Percentage of free space.
  • NAS Filer NFS Exported File System: Percentage of free space.
  • Process Families: Number of open files, total number of all related connections, total count of process family member connections.

The heuristics applied to certain attributes in scan results during a comparison are listed in the following table.

The following variables are used in the expressions:

  • X = The maximum value of the attribute between the two scan results.
  • N = The minimum value of the attribute between the two scan results.
  • P = The percentage change in value of the attribute between scan results.

Object Attribute

Equation

server — 15 minutes load average

X > 0.8 * cpu count AND P > 20%

OR

X > max(1, 0.25 * cpu count) AND P > 100%

server — percentage memory free (%)

N < 0.1 * total mem AND P > 25%

filesystem — percentage free (%)

N < 0.2 * size AND P > 10%

process family — open file count (on any member process)

X > 50 AND P > 50%

process family — connection count (aggregate across all member processes)

X > 50 * process count AND P > 30%

process family - connection count (on any member process)

X > 50 AND P > 50%